Urls.php

Go to the documentation of this file.

<?php

namespace Elgg\Http;

class Urls {

    public function addQueryElementsToUrl(string $url, array $elements): string {
        $url_array = parse_url($url);

        if (isset($url_array['query'])) {
            $query = elgg_parse_str($url_array['query']);
        } else {
            $query = [];
        }
    
        foreach ($elements as $k => $v) {
            if ($v === null) {
                unset($query[$k]);
            } else {
                $query[$k] = $v;
            }
        }
    
        // why check path? A: if no path, this may be a relative URL like "?foo=1". In this case,
        // the output "" would be interpreted the current URL, so in this case we *must* set
        // a query to make sure elements are removed.
        if ($query || empty($url_array['path'])) {
            $url_array['query'] = http_build_query($query);
        } else {
            unset($url_array['query']);
        }
        
        $string = $this->buildUrl($url_array, false);
    
        // Restore relative protocol to url if missing and is provided as part of the initial url (see #9874)
        if (!isset($url_array['scheme']) && (str_starts_with($url, '//'))) {
            $string = "//{$string}";
        }
        
        return $string;
    }

    public function addActionTokensToUrl(string $url, bool $html_encode = false): string {
        $url = $this->normalizeUrl($url);
        $components = parse_url($url);
    
        if (isset($components['query'])) {
            $query = elgg_parse_str($components['query']);
        } else {
            $query = [];
        }
    
        if (isset($query['__elgg_ts'], $query['__elgg_token'])) {
            return $url;
        }
    
        // append action tokens to the existing query
        // CSRF service is not DI injected because Urls is used by installer and CSRF requires DB installed
        $query['__elgg_ts'] = _elgg_services()->csrf->getCurrentTime()->getTimestamp();
        $query['__elgg_token'] = _elgg_services()->csrf->generateActionToken($query['__elgg_ts']);
        $components['query'] = http_build_query($query);
    
        // rebuild the full url
        return $this->buildUrl($components, $html_encode);
    }
    
    public function buildUrl(array $parts, bool $html_encode = true): string {
        // build only what's given to us
        $scheme = isset($parts['scheme']) ? "{$parts['scheme']}://" : '';
        $host = isset($parts['host']) ? "{$parts['host']}" : '';
        $port = isset($parts['port']) ? ":{$parts['port']}" : '';
        $path = isset($parts['path']) ? "{$parts['path']}" : '';
        $query = isset($parts['query']) ? "?{$parts['query']}" : '';
        $fragment = isset($parts['fragment']) ? "#{$parts['fragment']}" : '';
    
        $string = $scheme . $host . $port . $path . $query . $fragment;
    
        return $html_encode ? htmlspecialchars($string, ENT_QUOTES, 'UTF-8', false) : $string;
    }
    
    public function normalizeUrl(string $url): string {
        $url = str_replace(' ', '%20', $url);
    
        if ($this->isValidMultiByteUrl($url)) {
            // fix invalid scheme in site url
            $protocol_less_site_url = preg_replace('/^https?:/i', ':', elgg_get_site_url());
            $protocol_less_site_url = rtrim($protocol_less_site_url, '/');
            $protocol_less_site_url = str_replace('/', '\/', $protocol_less_site_url);
    
            return preg_replace("/^https?{$protocol_less_site_url}\/?/i", elgg_get_site_url(), $url);
        }
    
        $matches = [];
        if (preg_match('#^([a-z]+)\\:#', $url, $matches)) {
            // we don't let http/https: URLs fail filter_var(), but anything else starting with a protocol
            // is OK
            if ($matches[1] !== 'http' && $matches[1] !== 'https') {
                return $url;
            }
        }
    
        if (preg_match('#^(\\#|\\?|//)#', $url)) {
            // starts with '//' (protocol-relative link), query, or fragment
            return $url;
        }
    
        if (preg_match('#^[^/]*\\.php(\\?.*)?$#', $url)) {
            // root PHP scripts: 'install.php', 'install.php?step=step'. We don't want to confuse these
            // for domain names.
            return elgg_get_site_url() . $url;
        }
    
        if (preg_match('#^[^/?]*\\.#', $url)) {
            // URLs starting with domain: 'example.com', 'example.com/subpage'
            return "http://{$url}";
        }
    
        // 'page/handler', 'mod/plugin/file.php'
        // trim off any leading / because the site URL is stored
        // with a trailing /
        return elgg_get_site_url() . ltrim($url, '/');
    }
    
    public function isUrlIdentical(string $url1, string $url2, array $ignore_params): bool {
        $url1 = $this->normalizeUrl($url1);
        $url2 = $this->normalizeUrl($url2);
    
        if ($url1 === $url2) {
            return true;
        }
    
        $url1_info = parse_url($url1);
        $url2_info = parse_url($url2);
    
        if (isset($url1_info['path'])) {
            $url1_info['path'] = trim($url1_info['path'], '/');
        }
        
        if (isset($url2_info['path'])) {
            $url2_info['path'] = trim($url2_info['path'], '/');
        }
    
        // compare basic bits
        $parts = ['scheme', 'host', 'path'];
    
        foreach ($parts as $part) {
            if (isset($url1_info[$part], $url2_info[$part]) && $url1_info[$part] !== $url2_info[$part]) {
                return false;
            } elseif (isset($url1_info[$part]) && !isset($url2_info[$part])) {
                return false;
            } elseif (!isset($url1_info[$part]) && isset($url2_info[$part])) {
                return false;
            }
        }
    
        // quick compare of get params
        if (isset($url1_info['query'], $url2_info['query']) && $url1_info['query'] === $url2_info['query']) {
            return true;
        }
    
        // compare get params that might be out of order
        $url1_params = [];
        $url2_params = [];
    
        if (isset($url1_info['query'])) {
            $url1_info['query'] = html_entity_decode($url1_info['query']);
            if (!elgg_is_empty($url1_info['query'])) {
                $url1_params = elgg_parse_str($url1_info['query']);
            }
        }
    
        if (isset($url2_info['query'])) {
            $url2_info['query'] = html_entity_decode($url2_info['query']);
            if (!elgg_is_empty($url2_info['query'])) {
                $url2_params = elgg_parse_str($url2_info['query']);
            }
        }
    
        // drop ignored params
        foreach ($ignore_params as $param) {
            unset($url1_params[$param]);
            unset($url2_params[$param]);
        }
    
        // array_diff_assoc only returns the items in arr1 that aren't in arrN
        // but not the items that ARE in arrN but NOT in arr1
        // if arr1 is an empty array, this function will return 0 no matter what.
        // since we only care if they're different and not how different,
        // add the results together to get a non-zero (ie, different) result
        $diff_count = count($this->arrayDiffAssocRecursive($url1_params, $url2_params));
        $diff_count += count($this->arrayDiffAssocRecursive($url2_params, $url1_params));
        if ($diff_count > 0) {
            return false;
        }
    
        return true;
    }
    
    public static function isValidMultiByteUrl(string $url): bool {
        // based on http://php.net/manual/en/function.filter-var.php#104160
        if (filter_var($url, FILTER_VALIDATE_URL) !== false) {
            return true;
        }
    
        // Check if it has unicode chars.
        $l = elgg_strlen($url);
        if (strlen($url) === $l) {
            return false;
        }
    
        // Replace wide chars by X
        $s = '';
        for ($i = 0; $i < $l; ++$i) {
            $ch = elgg_substr($url, $i, 1);
            $s .= (strlen($ch) > 1) ? 'X' : $ch;
        }
    
        // Re-check now.
        return (bool) filter_var($s, FILTER_VALIDATE_URL);
    }
    
    protected function arrayDiffAssocRecursive(): array {
        $args = func_get_args();
        $diff = [];
        
        foreach (array_shift($args) as $key => $val) {
            for ($i = 0, $j = 0, $tmp = [$val], $count = count($args); $i < $count; $i++) {
                if (is_array($val)) {
                    if (empty($args[$i][$key]) || !is_array($args[$i][$key])) {
                        $j++;
                    } else {
                        $tmp[] = $args[$i][$key];
                    }
                } elseif (!array_key_exists($key, $args[$i]) || $args[$i][$key] !== $val) {
                    $j++;
                }
            }
            
            if (is_array($val)) {
                $tmp = call_user_func_array([$this, 'arrayDiffAssocRecursive'], $tmp);
                if (!empty($tmp)) {
                    $diff[$key] = $tmp;
                } elseif ($j == $count) {
                    $diff[$key] = $val;
                }
            } elseif ($j == $count && $count) {
                $diff[$key] = $val;
            }
        }
        
        return $diff;
    }
}